Customer Segmentation Group work

Instructions

Imagine you and your team mates have just founded a Data Science start up. Your first customer is a fundraising organization asking for support with regard to making most of their donor and transaction data.
The Head of Fund Development who is your direct counterpart and project sponsor asks your team to come up with a segmentation of the donor base as quickly as possible. The manager has a basic understanding of RFM modelling and tells you that there was a simple model in use some years ago which was neither updated nor further developed.
In general, your client organization is model-agnostic and trusts your judgment and consulting. However, the let you know that seeing alternative approaches in action together with your reflection, expertise and concluding recommendation how to process would be great …
The clients finally provide you with a flat file together with a list of feature descriptions …

Tasks

Form a group of 3 to 4 persons
Get accustomed to the data and have an explorative look at it.
Think your variables that might be added to or derived from the dataset with relative ease, e.g. conducting some research, data enrichment etc.
Take those down and formulate recommendations towards the client.
If you find time and a viable data source, you may of course go ahead and enrich the dataset
Apply at least two customer segmentation approaches to the provided data
Model examples: RFM, k-Means-Algorithm
Summarize the segmentation generated results and derived insights
Compare the model outputs and formulate a recommendation for the customer

Expected Output

Deliverable: Pitch presentation
Deadline: December 17th, 2021

Prepro

library(tidyverse)

## -- Attaching packages --------------------------------------- tidyverse 1.3.1 --

## v ggplot2 3.3.5     v purrr   0.3.4
## v tibble  3.1.6     v dplyr   1.0.7
## v tidyr   1.1.4     v stringr 1.4.0
## v readr   2.1.1     v forcats 0.5.1

## -- Conflicts ------------------------------------------ tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

feature_description_original <- readxl::read_excel(
  "data/feature_description.xlsx")
feature_description_original

customer_segmentation_raw <- read_csv2(
  "data/customer_segmentation_test.csv",
  col_types = list(col_character(), col_character(), col_character(), col_character(),
                col_double(), col_double(), col_character(), col_double(), col_double(),
                col_character(), col_double(), col_double(), col_character(), col_double(),
                col_double(), col_character(), col_double(), col_double(), col_character(),
                col_character(), col_character()),
  guess_max = 400000
) %>% mutate(
  `Date of Birth` = lubridate::dmy(`Date of Birth`),
  Gender = as.factor(Gender),
  MERCHANDISE2015 = as.factor(MERCHANDISE2015),
  MERCHANDISE2016 = as.factor(MERCHANDISE2016),
  MERCHANDISE2017 = as.factor(MERCHANDIESE2017),
  MERCHANDISE2018 = as.factor(MERCHANDIESE2018),
  MERCHANDISE2019 = as.factor(MERCHANDISE2019),
  LastPaymentDate = lubridate::dmy(LastPaymentDate),
  PenultimatePaymentDate = lubridate::dmy(PenultimatePaymentDate)
) %>% select(-c(MERCHANDIESE2017, MERCHANDIESE2018)) %>%
  rename(DateOfBirth = `Date of Birth`,
         ID =`Customer Number`)

## i Using "','" as decimal and "'.'" as grouping mark. Use `read_delim()` for more control.

skimr::skim(customer_segmentation_raw)

Data summary
Name	customer_segmentation_raw
Number of rows	406734
Number of columns	21
_______________________
Column type frequency:
character	2
Date	3
factor	6
numeric	10
________________________
Group variables	None

Variable type: character

skim_variable	n_missing	complete_rate	min	max	empty	n_unique	whitespace
ID	0	1.00	10	10	0	406734	0
Postcode	9176	0.98	1	9	0	2982	0

Variable type: Date

skim_variable	n_missing	complete_rate	min	max	median	n_unique
DateOfBirth	155491	0.62	1902-04-21	2015-03-30	1948-03-09	25514
LastPaymentDate	0	1.00	2015-01-03	2020-02-13	2018-12-06	1361
PenultimatePaymentDate	44699	0.89	1995-12-31	2020-02-05	2017-04-12	5376

Variable type: factor

skim_variable	complete_rate	ordered	n_unique	top_counts
Gender	1	FALSE	3	fem: 203904, mal: 183467, fam: 19363
MERCHANDISE2015	1	FALSE	2	0: 401845, 1: 4889
MERCHANDISE2016	1	FALSE	2	0: 401585, 1: 5149
MERCHANDISE2019	1	FALSE	2	0: 401470, 1: 5264
MERCHANDISE2017	1	FALSE	2	0: 402378, 1: 4356
MERCHANDISE2018	1	FALSE	2	0: 401470, 1: 5264

Variable type: numeric

skim_variable	complete_rate	mean	sd	p50	p75	p100	hist
COUNT2015	1	2.52	4.00	2	2	96.0	▇▁▁▁▁
SUM2015	1	42.44	850.19	15	45	388113.6	▇▁▁▁▁
COUNT2016	1	1.22	2.02	1	1	178.0	▇▁▁▁▁
SUM2016	1	50.93	591.05	16	50	295599.8	▇▁▁▁▁
COUNT2017	1	1.06	1.91	0	1	95.0	▇▁▁▁▁
SUM2017	1	24.78	572.90	0	20	207134.7	▇▁▁▁▁
COUNT2018	1	1.00	1.87	0	1	49.0	▇▁▁▁▁
SUM2018	1	20.64	1552.60	0	15	911146.5	▇▁▁▁▁
COUNT2019	1	0.97	1.79	0	1	31.0	▇▁▁▁▁
SUM2019	1	46.44	3999.80	0	30	2400000.0	▇▁▁▁▁

feature engineering

Bin hier sehr offen für Verbesserungsvorschläge ^^

zip_code_list <- readxl::read_excel("data/PLZ_Verzeichnis-20211201.xls")
zip_code_list

customer_segmentation_with_zip <- customer_segmentation_raw %>%
  left_join(zip_code_list, by = c("Postcode" = "PLZ")) %>%
  select(-c(`gültig ab`, `gültig bis`, NamePLZTyp, intern_extern, adressierbar, Postfach)) %>%
  drop_na(Postcode, Ort, Bundesland) %>%
  mutate(Postcode = as.factor(Postcode),
         Bundesland = as.factor(Bundesland))

customer_segmentation_with_zip

# here we define, which months should be understood as "christmas months" to define "XMAS_donation"
XMAS_months = c(11,
                12,
                1)

# this date will be used as the reference for this analysis
reference_date <- lubridate::ymd("2021-12-17")

customer_segmentation_first_prepro <- customer_segmentation_with_zip %>%
  mutate(
    # year of customer's birthday
    year_born = lubridate::year(DateOfBirth),
    
    # age of donors at their last donation
    age_at_last_donation = lubridate::interval(DateOfBirth, LastPaymentDate) %>%
      as.numeric("years") %>%
      as.integer(),
    
    generation_moniker = case_when(
      year_born <= 1945 ~ "silent" ,
      year_born <= 1964 ~ "boomer",
      year_born <= 1980 ~ "x",
      year_born <= 1996 ~ "millennial",
      year_born <= 2012 ~ "z"
    ) %>% as_factor(),

    # total number of donations over all years
    COUNTtotal = COUNT2015+
                 COUNT2016+
                 COUNT2017+
                 COUNT2018+
                 COUNT2019,

    # total donation amount over all years
    SUMtotal = SUM2015+
               SUM2016+
               SUM2017+
               SUM2018+
               SUM2019,

    # average donation amount
    SUMaverage = SUMtotal / COUNTtotal,

    # month of the last payment
    LastPaymentMONTH = lubridate::month(LastPaymentDate) %>% as.factor(),

    # month of second to last payment
    PenultimatePaymentMONTH = lubridate::month(PenultimatePaymentDate) %>% as.factor(),

    # year of the last payment
    LastPaymentYEAR = lubridate::year(LastPaymentDate),

    # year of second to last payment
    PenultimatePaymentYEAR = lubridate::year(PenultimatePaymentDate),

    # THIS ONE NEEDS WORK
    # status as christmas donor if the last two payments were around christmas,
    # but we have to tweak the time interval (is Nov to Jan too large?)
    # also: what about people that only have one payment in total, that should be considered. The "maybe" status is shady at best
    XMAS_donor = as_factor(case_when(LastPaymentMONTH %in% XMAS_months & PenultimatePaymentMONTH %in% XMAS_months ~ "yes",
                                     LastPaymentMONTH %in% XMAS_months ~ "maybe",
                                     TRUE ~ "unlikely")),

    # days between last and second to last payment
    donation_interval = lubridate::day(lubridate::days(LastPaymentDate - PenultimatePaymentDate)),
    
    # days since the last payment in relation to our reference date
    days_since_last_payment = as.integer(LastPaymentDate - reference_date),

    # binary factor variable expressing if any merchandise was bought over the observation period (clumsily coded)
    merchandise_any = as_factor(if_else(
                                  !is.na(MERCHANDISE2015) & MERCHANDISE2015 != 0 |
                                  !is.na(MERCHANDISE2016) & MERCHANDISE2016 != 0 |
                                  !is.na(MERCHANDISE2017) & MERCHANDISE2017 != 0 |
                                  !is.na(MERCHANDISE2018) & MERCHANDISE2018 != 0 |
                                  !is.na(MERCHANDISE2019) & MERCHANDISE2019 != 0,
                                  1,
                                  0))) %>%

  # grouping for the next mutation (num_of_donation_years)
  group_by(ID) %>%

  # number of years in which anything was donated (0-5)
  mutate(num_of_donation_years = sum(COUNT2015 > 0,
                                     COUNT2016 > 0,
                                     COUNT2017 > 0,
                                     COUNT2018 > 0,
                                     COUNT2019 > 0, na.rm=T)) %>%

  # ungrouping is important! ;)
  # I learned that skimr tries to show its output based on groups if working with a grouped dataset... that crashed my computer twice ^^
  ungroup() %>%

  # remove variables that have no further use or
  select(-c(ID, DateOfBirth, LastPaymentDate, PenultimatePaymentDate))

customer_segmentation_first_prepro

customer_segmentation_first_prepro %>% skimr::skim()

Data summary
Name	Piped data
Number of rows	396694
Number of columns	34
_______________________
Column type frequency:
character	1
factor	13
numeric	20
________________________
Group variables	None

Variable type: character

skim_variable	n_missing	complete_rate	min	max	empty	n_unique	whitespace
Ort	0	1	2	40	0	2178	0

Variable type: factor

skim_variable	n_missing	complete_rate	ordered	n_unique	top_counts
Gender	0	1.00	FALSE	3	fem: 199545, mal: 179215, fam: 17934
Postcode	0	1.00	FALSE	2249	122: 6776, 121: 6208, 110: 5941, 502: 5383
MERCHANDISE2015	0	1.00	FALSE	2	0: 391818, 1: 4876
MERCHANDISE2016	0	1.00	FALSE	2	0: 391552, 1: 5142
MERCHANDISE2019	0	1.00	FALSE	2	0: 391460, 1: 5234
MERCHANDISE2017	0	1.00	FALSE	2	0: 392339, 1: 4355
MERCHANDISE2018	0	1.00	FALSE	2	0: 391460, 1: 5234
Bundesland	0	1.00	FALSE	9	N: 88175, W: 70706, O: 66082, St: 57348
generation_moniker	146208	0.63	FALSE	5	sil: 110508, boo: 102068, x: 33020, mil: 4734
LastPaymentMONTH	0	1.00	FALSE	12	12: 119035, 11: 66379, 1: 45775, 10: 42275
PenultimatePaymentMONTH	37875	0.90	FALSE	12	12: 91203, 11: 56900, 10: 42674, 1: 27463
XMAS_donor	0	1.00	FALSE	3	unl: 165505, may: 119746, yes: 111443
merchandise_any	0	1.00	FALSE	2	0: 377620, 1: 19074

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
COUNT2015	0	1.00	2.56	4.03	0.00	0.00	2.00	4.00	96.0	▇▁▁▁▁
SUM2015	0	1.00	41.12	724.36	0.00	0.00	15.00	45.00	388113.6	▇▁▁▁▁
COUNT2016	0	1.00	1.24	2.03	0.00	0.00	1.00	1.00	178.0	▇▁▁▁▁
SUM2016	0	1.00	51.20	596.95	0.00	0.00	20.00	50.00	295599.8	▇▁▁▁▁
COUNT2017	0	1.00	1.08	1.92	0.00	0.00	0.00	1.00	95.0	▇▁▁▁▁
SUM2017	0	1.00	24.45	484.85	0.00	0.00	0.00	20.00	207134.7	▇▁▁▁▁
COUNT2018	0	1.00	1.02	1.88	0.00	0.00	0.00	1.00	49.0	▇▁▁▁▁
SUM2018	0	1.00	20.76	1570.91	0.00	0.00	0.00	15.00	911146.5	▇▁▁▁▁
COUNT2019	0	1.00	0.98	1.80	0.00	0.00	0.00	1.00	31.0	▇▁▁▁▁
SUM2019	0	1.00	46.90	4049.95	0.00	0.00	0.00	30.00	2400000.0	▇▁▁▁▁
year_born	146204	0.63	1949.25	14.01	1902.00	1939.00	1948.00	1959.00	2015.0	▁▇▇▂▁
age_at_last_donation	146204	0.63	68.33	14.00	0.00	59.00	70.00	79.00	117.0	▁▁▇▇▁
COUNTtotal	0	1.00	6.87	9.93	1.00	2.00	3.00	7.00	273.0	▇▁▁▁▁
SUMtotal	0	1.00	184.43	4898.70	0.01	30.00	65.00	160.00	2400225.0	▇▁▁▁▁
SUMaverage	0	1.00	36.08	1530.61	0.01	11.25	17.34	29.42	750000.0	▇▁▁▁▁
LastPaymentYEAR	0	1.00	2017.78	1.53	2015.00	2016.00	2018.00	2019.00	2020.0	▅▂▃▇▂
PenultimatePaymentYEAR	37875	0.90	2015.72	3.91	1995.00	2015.00	2017.00	2018.00	2020.0	▁▁▁▃▇
donation_interval	37875	0.90	773.66	1215.88	1.00	123.00	354.00	762.00	8766.0	▇▁▁▁▁
days_since_last_payment	0	1.00	-1293.24	561.24	-2540.00	-1814.00	-1102.00	-762.00	-673.0	▂▂▂▃▇
num_of_donation_years	0	1.00	2.50	1.49	1.00	1.00	2.00	4.00	5.0	▇▅▃▂▃

#Maybe it's a good idea to take out all the NAs for age. Obviously we lose a lot of rows, but 251000 left still seems plenty to me.
customer_segmentation_complete <- customer_segmentation_first_prepro %>% drop_na(year_born)
customer_segmentation_complete

Visual Exploration

ggplot(customer_segmentation_first_prepro, aes(XMAS_donor)) +
  geom_bar() +
  facet_wrap(~Gender)

ggplot(customer_segmentation_first_prepro, aes(num_of_donation_years)) +
  geom_bar() +
  facet_wrap(~generation_moniker)

ggplot(customer_segmentation_first_prepro %>% drop_na(age_at_last_donation), aes(age_at_last_donation)) +
  geom_histogram(binwidth = 5)

ggplot(customer_segmentation_first_prepro %>% filter(SUMtotal > 0 & SUMtotal < 5000), aes(x = SUMtotal)) +
  geom_histogram(binwidth = 100) +
  facet_wrap(~Gender)

ggplot(customer_segmentation_first_prepro, aes(LastPaymentMONTH)) +
  geom_bar() +
  facet_wrap(~Gender)

ggplot(customer_segmentation_first_prepro, aes(PenultimatePaymentMONTH)) +
  geom_bar() +
  facet_wrap(~Gender)

ggplot(customer_segmentation_first_prepro %>% filter(COUNTtotal < (7 * 6)), aes(COUNTtotal)) +
  geom_histogram(binwidth = 1)

ggplot(customer_segmentation_first_prepro %>% drop_na(donation_interval) %>% filter(donation_interval < (360 * 5)), aes(donation_interval)) +
  geom_histogram(binwidth = 30)

mean_total_sum <- customer_segmentation_first_prepro$SUMtotal %>% mean(na.rm = TRUE)
sd_total_sum <- customer_segmentation_first_prepro$SUMtotal %>% sd(na.rm = TRUE)

ggplot(customer_segmentation_first_prepro %>% drop_na(year_born) %>% filter(SUMtotal < (mean_total_sum + sd_total_sum * 6)), aes(year_born, SUMtotal)) +
  geom_point(alpha = 1 / 10)

# taken from https://de.statista.com/statistik/daten/studie/75396/umfrage/entwicklung-der-bevoelkerung-in-oesterreich-nach-bundesland-seit-1996/
pop_vienna <- 1921153
pop_lower_austria <- 1691040
pop_upper_austria <- 1495756
pop_styria <- 1247159
pop_tyrol <- 760161
pop_carithia <- 562230
pop_salzburg <- 560643
pop_vorarlberg <- 399164
pop_burgenland <- 296040

donors_per_state_per_100_000_inhabitants <- customer_segmentation_first_prepro %>%
  select(Bundesland) %>%
  group_by(Bundesland) %>%
  count() %>%
  ungroup() %>%
  mutate(
    n = case_when(
      Bundesland == "B" ~ n / pop_burgenland * 100000,
      Bundesland == "K" ~ n / pop_carithia * 100000,
      Bundesland == "N" ~ n / pop_lower_austria * 100000,
      Bundesland == "O" ~ n / pop_upper_austria * 100000,
      Bundesland == "Sa" ~ n / pop_salzburg * 100000,
      Bundesland == "St" ~ n / pop_styria * 100000,
      Bundesland == "T" ~ n / pop_tyrol * 100000,
      Bundesland == "V" ~ n / pop_vorarlberg * 100000,
      Bundesland == "W" ~ n / pop_vienna * 100000
    )
  )

ggplot(donors_per_state_per_100_000_inhabitants, aes(Bundesland, n)) +
  geom_col()

ggplot(customer_segmentation_first_prepro, aes(days_since_last_payment)) +
  geom_histogram(binwidth = 30)

RFM

RFM segments customers according to three variabless: Recency, Frequency, Monetary Value. Using the rfm package, RFM scores can be computed either on raw transaction data (one row per transaction), or on aggregated customer data (one row per customer). For the former, the function rfm_table_order can be used, for the latter either rfm_table_customer or rfm_table_customer2. Since our dataset represents aggregated customer data, the latter should be used. It can be computer directly from the raw data:

library(rfm)

rfm_scores <- customer_segmentation_raw %>%
  # create new variables: total donation sum; total number of donations
  mutate(SUMtotal = SUM2015 + SUM2016 + SUM2017 + SUM2018 + SUM2019,
         COUNTtotal = COUNT2015 + COUNT2016 + COUNT2017 + COUNT2018 + COUNT2019,
         LastPaymentDate = as.Date(LastPaymentDate)) %>%
  # compute RFM scores
  rfm_table_customer_2(customer_id = ID,
                       n_transactions = COUNTtotal,
                       latest_visit_date = LastPaymentDate,
                       total_revenue = SUMtotal,
                       analysis_date = reference_date)

rfm_scores

## Warning in `[<-.data.frame`(`*tmp*`, is_list, value = list(`1` =
## "<tibble[,8]>", : replacement element 1 has 1 row to replace 0 rows

## Warning in `[<-.data.frame`(`*tmp*`, is_list, value = list(`1` =
## "<tibble[,8]>", : replacement element 2 has 1 row to replace 0 rows

Visual inspection of RFM scores:

We can see that higher monetary values are characterized by higher donation frequencies and more recent donations. There is an obvious cluster of low monetary value for frequency values in [1,2] and recency in [1,3]. These might be ‘sleepers’, i.e. customers who donated only a few times and are not active donors any more. In the upper left corner, we see very unrecent customers who donated above average for this recency score. It might be wort focusing on them, since they could be “reactivated” as donors, since they showed above-average donation willingness among low-frequency donors. Note: The higher the recency score, the more recent the last transaction!

rfm_heatmap(rfm_scores)

rfm_bar_chart(rfm_scores)

In the frequency vs monetary value plot below there is no strong correlation between donation frequency and monetary value. However, in the low-frequency area, there seems to be a slight positive correlation with monetary value.

rfm_fm_plot(rfm_scores)

R Notebook

Customer Segmentation Group work

Instructions

Tasks

Expected Output

Prepro

feature engineering

Visual Exploration

RFM